I have certain doubts about data cleaning in PI-System, it is an interesting problem to solve, let me know if anyone can help here.

I have setup a plant where some of the data issues are:

 

  • Sensor data lost due to connection for some time (5% of total data).
  • Some times data comes as "0" which is again not correct.
  • Some times garbage value might also come.

 

How does PI-System help in cleaning such data and if it does, then what method do they use (For example- Back calculation, interpolation or any other algorithm). Because if we perform algorithms of Root cause analysis or anomaly detection, then such data is not reliable.

 

If PI-system does not do that, how should I go ahead with manually clean the data before loading it up for different analyses.